Project Grant 2214925

This $1,197,878 project grant from the National Science Foundation's Office of Advanced Cyberinfrastructure will support the development of Evolutional Deep Neural Network algorithms for solving high-dimensional partial differential equations. Funded under the Computer and Information Science and Engineering program (CFDA 47.070), this collaboration between U.S. and French researchers aims to accelerate computational predictions of complex phenomena across multiple disciplines. Specifically, the researchers at The Johns Hopkins University will develop fundamental techniques to optimize network architecture design, enable dynamic adaptivity of solutions, and scale algorithms for massive parallelism. Evaluation against benchmark data will assess performance. This work seeks to advance the use of machine learning for computationally modeling systems governed by high-dimensional equations, with applications in engineering, physics, medicine, and public health.